Five of 22 accounts opened cold. What they did differently.

Of the 22 DTC ad accounts I read in the Meta Ad Library on July 17, 2026, five had a genuine cold-audience entry: an ad that named a problem before it named a product, running from the brand's own page, not as a single line buried under a product claim. Waterboy, Maev, Diamond Brew, Bushbalm, and Jones Road Beauty. They are in five different categories and they share three things.

The first is that the problem is stated as a fact about the world, not as a question about the product. Waterboy's line about other hydration brands being sugar water in disguise is an argument someone can agree with before they have heard of Waterboy. Maev opens with what if your dog's food met the same standards as yours — a sentence that works on a person who has never considered raw feeding. Neither ad needs the brand to exist in order to make sense. That is the actual test, and most top-of-funnel creative fails it: take the brand name out, and if the sentence collapses, it was never speaking to a stranger.

The second is that the stages are covered by different ads, not by one ad trying to do everything. Diamond Brew had 22 own-page ads split across four jobs: a testimonial calling it the GOAT of instant coffee, a product spec at 160 milligrams ready in seconds, problem-aware lines about bitterness and about being clean and third-party tested, and a ten-percent offer. Four stages, four sets of creative, one account. The accounts that failed usually had the opposite shape — many ads, one job. Obvi's six ads were six executions of a single angle.

The third is the least visible and probably the most important: these accounts had somewhere for a cold reader to land that was not a product page. Bushbalm called out ingrown hairs and razor burn at the top and ran a separate business line for estheticians underneath. Jones Road opened on foundation sticks going flat and cakey and heavy, then moved into a quiz. A quiz is not a clever conversion trick; it is an answer to the problem the ad just raised, for someone who does not yet know which product is theirs. Send that person to a product grid and the ad was wasted regardless of how good the hook was.

None of this is expensive. Every one of these is a copy decision and a landing decision. There is no new asset class here, no production budget, no channel to open. The five accounts that opened cold were not spending more than the seventeen that did not — I have no spend data for any of them, so I cannot say that either way, but I can say that nothing I saw required money the others do not have.

It is worth saying what the five did not have in common, because it rules out the easy explanations. They were not the biggest: Ridge is far larger than Waterboy and covered one stage. They were not the youngest or the oldest. They were not concentrated in one category — electrolytes, dog food, coffee, body care, makeup. They did not share an agency as far as I can tell. Whatever produced the difference, it was not scale, sector, or vintage.

If you want to apply this to your own account, the three checks map directly. Take your top-performing ad and delete every mention of your brand: does the first sentence still say something true and interesting? Count how many distinct jobs your live ads do, as opposed to how many ads you have. And open the destination of your coldest ad and ask whether someone who just learned their problem has a reason to be on that page.

The limits are the same ones I stated in the other two posts and they matter more here, because this one names brands approvingly. The Ad Library shows ads active at the moment of viewing; this is July 17 and nothing more. The top view can be partial, so an account I called narrow may have had more running than I saw. And I am reading creative, not results — I do not know whether any of this worked for any of them. What I can show is what was live, and what the five had in common that the other seventeen did not.